JPH08272806A - Data base retrieval system - Google Patents

Abstract

PURPOSE: To automatically judge the conditions of a retrieval formula inputted by a retrieving person, to utilize the merits of various retrieval methods and to perform efficient retrieval in a data base retrieval system capable of executing the plural retrieval methods. CONSTITUTION: As a retrieval means for retrieving a data base inside a data base storage part 11, for instance, an index retrieval means 13 and an entire sentence retrieval means 14 are provided. Then, this system is provided with an input means 15 for inputting the retrieval formula, a division means 21 for dividing the retrieval formula into monomials, an allocation means 32 for allocating the monomials obtained in the division means 21 respectively to the index retrieval means 13 or the entire sentence retrieval means 14 as retrieval keys and an arithmetic means 33 for performing a logic operation based on the retrieval formula for a retrieved result from the index retrieval means 13 and the entire sentence retrieval means 14 and performing output to a display means 16 as the retrieved result corresponding to the retrieval formula.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a database search system for searching a document database or the like, and more particularly to a database search system in which a plurality of search engines by different search methods are mounted on the same device.

[0002]

2. Description of the Related Art A database search system selects information required by a searcher from a large amount of information by using a search keyword or a search expression combining them. Typically, an extremely large number of documents such as patent gazettes, academic papers, newspaper articles, etc. are stored as primary information in a document database, and documents corresponding to the search formula conditions entered by the searcher are stored in this document database. It is built as a system to find out from inside. There are various kinds of information detected by the database search system, and there are some search methods, such as those that emphasize search speed and search accuracy, depending on the file management method and algorithm. Each has its strengths and weaknesses.

Examples of such a search method include an index search method and a full-text search method. The index search method is to extract the bibliographical information and assigned classifications from the primary information to make it secondary information, and reorganize the secondary information to provide an index, which corresponds to the input search keyword. This is a method of detecting whether or not there is something to do in the index, and trying to access the original primary information according to the result. Since this method only needs to search the index, it is suitable for performing a high-speed search for a large number of documents, but it cannot search for items that are not in the index. On the other hand, the full-text search method is a method in which an entire file of primary information is searched to find a document including a character string corresponding to a search keyword. The full-text search method can use any search keyword and can perform a detailed search. However, since the entire primary information file is the search target, a large amount of computer resources is required when the search target range is wide. It has the drawback of consuming too much.

Further, the database search system includes a plurality of search engines (search means) which have different search methods.
There are some with. In such a database search system, one of the search methods is selected according to the search condition to execute the search.

[0005]

In the above-mentioned conventional database search system, even when a plurality of search methods can be executed, any one search method is selected for the search process for each search expression. However, there is a problem that it is strongly affected by the drawbacks peculiar to the search method. For example, if you use a database search system that can perform both index search and full-text search, one of these methods is used for searching, so if you try to search a wide area in detail, you will sacrifice the detail. Either (in the case of index search) or consume a large amount of computer resources (full-text search).

It is an object of the present invention to automatically determine the conditions of a search expression input by a searcher in a database search system capable of executing a plurality of search methods and utilize the advantages of each search method to achieve efficient It is to provide a database search system capable of performing a search.

[0007]

A first database search system of the present invention comprises an input means for inputting a search expression, a dividing means for dividing the search expression into monomials, and a plurality of databases for searching by different search methods. Search method of
Assigning means for allocating the monomials to the corresponding corresponding search means, and operation for calculating results obtained by searching the database by each search means based on the search expressions and outputting as a search result corresponding to the search expressions Means for searching the database based on the unary expression assigned to the search means by the assigning means.

The second database search system of the present invention comprises an input means for inputting a search expression, a plurality of search means for searching a database by different search methods, and the search expression so that the search expression is adapted to each search method. Division and allocation means for allocating each of the partial search expressions to the corresponding search means by dividing the partial search expression, and a result obtained by the search means searching the database based on the search expression to perform the search. Calculating means for outputting a search result corresponding to the expression, and each of the searching means searches the database based on the partial search expression assigned to the searching means by the division assigning means.

In the present invention, search target range setting means may be provided to set a search target range common to each search means, and each search means may execute a search within this search target range. Further, as the plurality of search means, for example, a search means for performing an index search and a search means for performing a full-text search can be cited.

[0010]

The input search formula is divided to generate a unary formula or a partial search formula, and these unary formulas or partial search formulas are assigned to the corresponding search means. Therefore, each search means takes advantage of its advantage. The search is done. Then, the search result corresponding to the input search formula is finally obtained by performing a logical operation on the search result by each search means based on the search formula.

[0011]

Embodiments of the present invention will now be described with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a database search system according to an embodiment of the present invention. here,
The case of performing patent information search will be described as an example.

Literature information, which is primary information, is recorded in the database storage unit 11. Here, patents, utility models, and other public and official gazettes, utility models (for applications after January 1994), registered designs, etc. are computerized and recorded in the database storage unit 11. These publications are classified and stored in three types of database groups (hereinafter, abbreviated as DBG): open, public notice, and registration. In FIG. 1, these three DBGs are DBG # 1 to DBG # 1.
This is shown in DBG # 3. Although not shown, a DBG information file that describes what kind of DBG there is is also stored in the database storage unit 11.

Within each DBG, patents, utility models, designs,
A database (hereinafter abbreviated as DB) is provided according to the classification of international applications etc. according to the PCT (Patent Cooperation Treaty), and in FIG. 1, DBG # corresponding to the publication is disclosed.
1 is provided with four DBs (DB # 1 to DB # 4). In this embodiment, it is assumed that a unique number (hereinafter referred to as a document ID number) is given to each of the patent documents. This document ID number is, for example, Japanese Patent Publication No. 4-12.
For Japanese Patent No. 3456, "B404123456", Shokai 63-
For example, Japanese Patent No. 112233 has a designation such as "U363112233". Also, what DB is in the DBG
A DB information file (not shown) describing whether or not exists is also provided in the database storage unit 11.

Each DB has an index data section 21 which is an area in which data used for performing an index search is recorded, and a full text search data section which is an area in which data is used for performing a full text search. 22. In the case of the index search of the publication, the search is often performed without dividing the patent and the utility model. Therefore, in this embodiment, it is common to all cases in one DBG (DB # 1 to DB # 4 in this example). (Common to all), the index data section 21 is set. The index data section 21 includes items that are key when searching patent documents,
For example, an index file that includes the application date and publication date, the application number, the publication number, the applicant name, the granted international patent classification, etc. as an index is included, and an index search is performed using the index data section 21. The document ID number for the corresponding patent document can be acquired.

On the other hand, the full-text search data section 22 is further divided into a plurality of categories 23. Category 2
Reference numeral 3 is for dividing data used for performing full-text search on a monthly basis, for example, in order of publication date, and the data is stored in any of the categories 23. Here, the reason why the data is divided into categories 23 is that it is easier to handle if the data for all-retrieval search is divided into some units, and the search range is limited to a specific category. This is because the search time can be shortened. In the present embodiment, the data used to perform the full-text search may be the full-text data of each patent document (the data itself as the primary information), or the search method used for the full-text search. It may be data processed from full-text data according to the above. The full-text search is generally performed in order to know what the document ID number of a patent document containing a certain character string (search keyword) is, and where in the patent document the character string exists. Since it is often unnecessary to know the information, for example, as disclosed in Japanese Patent Laid-Open No. 4-326164, autocorrelation information is extracted from the full-text data of each patent document and used as full-text search data for each patent document. As a result, the storage capacity required for the full-text search data section 22 can be greatly reduced.

Now, in this database search system,
An index search unit 13 that accesses the index data unit 21 to perform an index search for a specified item in all cases in one DBG, and a full text to perform a full text search in a specified DB or category 23 in the DBG. The search means 14 and the index search means 13 and the full-text search means 14 are instructed about search targets and search keys, and the index search means 13 and the full-text search means 1 are specified.
4 is provided with a central processing unit 12 for processing the search results. Further, this database search system has an input unit 15 connected to the central processing unit 12 and including a keyboard and a mouse, and a display unit 16 connected to the central processing unit 12 such as a CRT. The index search unit 13 and the full-text search unit 14 are configured to return the document ID number of the property (patent document in this case) corresponding to the search key.

In the central processing unit 12, the dividing means 31 for dividing the search expression input from the input means 15 into monomials, and the monomials obtained by the dividing means 31 are either index search means 13 or full-text search means 14, respectively. To the search means from the index search means 13 and the full-text search means 14 based on the above-mentioned search formula, and output to the display means 16 as the search result corresponding to this search formula. Means 33
And a search target range setting means 34 for setting a search target range in the index search means 13 and the full text search means 14.

Next, the operation of this database search system will be described. FIG. 2 is a flowchart showing the procedure of the entire processing.

First, a search range is set (step 51). In the case of patent document search, database storage unit 11
The search is rarely carried out suddenly for all the cases stored in, and the search is limited to the publications or publications published within a certain period or limited to the publications within a certain period. It is common to do this. Therefore, in this embodiment, a process for limiting the search target range is executed prior to the actual search process. The detailed procedure of the search target range setting process is shown in FIG. First, the searcher selects one DBG using the input means 15 (step 6).
1). For example, the publication gazette DBG (DBG # 1 in FIG. 1) is selected. As a result, the DBG information file and the DB information file corresponding to this DBG are read, and the list of the DBs and categories included in the selected DBG is displayed on the screen of the display means 16 (step 62).

Next, the searcher selects a desired DB and category (step 63). For example, the patent DB and utility model DB are selected as DBs, and the category 199
Select the one released in January, 2014. The shaded category 23 in FIG. 1 is the selected category. Then, an inquiry is made as to whether another DB or category is to be further selected (step 64), and when additional selection is made, step 63 is performed.
Returning to step 2, when no additional selection is made, the search target range setting means 34 sets the search target range for each of the index search means 13 and the full-text search means 14. For the full-text search means 14, the search target range is set by designating which category is searched. Regarding the index search means 13, the search target range is set by giving a search key corresponding to the selected DB or category to the index search means 13. In this example, it is a patent or utility model,
A search key having a release date from January 1, 1994 to January 31, 1994 is given.

When the search target range setting process of step 51 is completed as described above, the search process is subsequently executed (step 52). FIG. 4 is a flowchart showing a specific procedure of the search process. In the search process, first, the searcher inputs a search expression using the input means 15 (step 7).
1). This search formula is accepted by the dividing means 31 and divided into monomials for each search key. Since it is necessary to execute a logical operation based on this search formula as described later, this search formula is also input to the calculation means 33. For example,
International patent classification is G06F15 / 40 or G06F15
If the user wants to search for documents that include the character string "display" or "display" in the specification, the search keys are the following formulas (1) to (4).

International Classification # 1 = G06F15 / 40 (1) International Classification # 2 = G06F15 / 60 (2) Keyword # 1 = "Display" ... (3) Keyword # 2 = "Display" (4) , The polynomial shown in Expression (5) is input as the search expression.

{Equation (1) or Equation (2)} and {Equation (3) or Equation (4)} (5) The dividing unit 31 inputs a search expression as shown in Equation (5), This is divided and the monomial expressions respectively expressed by equations (1) to (4) are output. Of these monomials, equation (1), equation
(2) is a search key suitable for the search by the index search means 13, and expressions (3) and (4) are search keys (search keywords) suitable for the search by the full-text search means 14.

These monomials obtained by the dividing means 31 are input to the allocating means 32. Assigning means 32
Assigns each monomial expression to the index search means 13 or the full-text search means 14, respectively, whichever is more suitable, and the index search means 13 and the full-text search means 14 execute a search based on the monomial expression assigned to each. In the above example, the monomial expressions represented by formulas (1) and (2) are assigned to the index search means 13, and the monomial expressions represented by formulas (3) and (4) are assigned to the full-text search means 14. . In the following, the procedure for unary expression assignment and the search by the index search means 13 and the full-text search means 14 will be described in detail.

The assigning means 32 first selects one from the input monomials and judges whether or not the monomials are the monomials for full-text search (step 73). If it is not a full-text search target, the process proceeds to step 76, and if it is a full-text search target, the monomial expression is assigned to the full-text search means 14. As a result, the full-text search means 14 performs a full-text search within the above-described search target range based on the monomials (step 74), obtains the document ID number of the relevant patent document (step 75), and moves to step 76. To do.

In step 76, it is judged whether the selected monomial expression is a monomial expression to be index-searched. If it is not the index search target, the process proceeds to step 79. If it is the index search target, the unary expression is assigned to the index search means 13. as a result,
The index search means 13 executes an index search within the above-described search target range based on the unary expression (step 77), and acquires the document ID number of the corresponding patent document (step 78). Then, in step 79, it is determined whether or not there remains unselected monomials. If it remains, the process returns to step 73 to perform the same processing for the next monomial expression, and if it does not remain, the calculation means 33.
Then, the process proceeds to step 80 in order to carry out the processing in.

In step 80, the arithmetic means 33 executes a logical operation based on the retrieval formula for the retrieval results obtained in the above-mentioned steps 75 and 77.
That is, the search results for each unary expression obtained by dividing the search expression are integrated in the search result by following the reverse process of dividing the search expression. Then, the result obtained by the logical operation is output as the final search result for the search expression (step 81). In the above example, the index search means 1 is used for the four monomial expressions represented by the expressions (1) to (4).
3 or 4 sets of search results by the full-text search means 14 are obtained, and each search result is represented as a set of document ID numbers corresponding to the corresponding monomials. Then, in the computing means 33, the retrieval formula, that is, the formula (5)
A logical operation is performed on the search result corresponding to each monomial expression according to the logical expression represented by, and the final search result is output as a set of document ID numbers of documents corresponding to the search expression.

When the search process of step 52 is executed as described above, the obtained search result is displayed on the display means 16 (step 53). In the above example, the patent application or utility model registration application published in January 1994, and the granted international patent classification is G06F15 / 40.
Alternatively, the publication number of the publication that is G06F15 / 60 and has the description of the character string "display" or "display" in the description is displayed. Then, an inquiry is made as to whether the processing is to be ended as it is (step 34).
Returning to, when a termination is instructed, a series of processing is terminated.

In the above embodiment, the search expression is divided into monomials and assigned to the index search means 13 and the full-text search means 14. However, in the present invention, it is not always necessary to divide them into monomials. In the above example, "equation (1) or
The logical expression (partial search expression) represented by the expression (2) is assigned to the index search means 13, and the index search means 1
3 performs an index search based on "expression (1) or expression (2)" to obtain search result # 1, and "expression (3) or expression"
(4) ”is assigned to the full-text search means 14, and the full-text search means 14 performs a full-text search based on“ expression (3) or expression (4) ”to obtain a search result # 2, The arithmetic means 33 may perform a logical operation of "search result # 1 and search result # 2". That is, a partial search expression suitable for the search means may be assigned to each search means. In this case, the dividing and assigning unit is configured by the dividing unit and the assigning unit.

Although the embodiments of the present invention have been described above, the plurality of search means in the present invention are not limited to the combination of the index search means and the full-text search means, and search means other than these search methods may be used. May be.

[0031]

As described above, according to the present invention, an input search expression is divided to generate a monomial expression or a partial search expression, and these monomial expression or partial search expression are assigned to the corresponding search means. By executing the search by the search means and performing the calculation based on the search formula on the search result in each search means, an efficient search can be performed in each search means by utilizing the merit of the search means. There is an effect. For example, by using a search unit that performs an index search and a search unit that performs a full-text search as the plurality of search units, it is possible to perform a search that achieves both high-speed index search and full-text search detail.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of a database search system according to an embodiment of the present invention.

FIG. 2 is a flowchart showing a procedure of overall processing.

FIG. 3 is a flowchart showing a procedure of search target range setting processing.

FIG. 4 is a flowchart showing a procedure of search processing.

[Explanation of symbols]

 11 database storage unit 12 central control unit 13 index search unit 14 full-text search unit 15 input unit 16 display unit 21 index data unit 22 full-text search data unit 23 category 31 dividing unit 32 assigning unit 33 computing unit 34 search range setting unit 51 ~ 54,61-65,71-81 steps

Claims (4)

[Claims] 1. A database search system for searching a database according to a search expression, comprising: input means for inputting the search expression; dividing means for dividing the search expression into monomials; and searching the database by different search methods. A plurality of search means, an assigning means for allocating the monomials to the corresponding search means, and a result obtained by searching the database by each search means, based on the search expression, corresponding to the search expression A database search system comprising: a calculation unit that outputs the search result, and each of the search units searches the database based on a monomial expression assigned to the search unit by the assigning unit. 2. A database search system for searching a database according to a search expression, comprising input means for inputting a search expression, a plurality of search means for searching a database by different search methods, and each search method. As described above, the search formula is divided into partial search formulas, and each of the partial search formulas is assigned to a corresponding search unit. And a calculation means for outputting as a search result corresponding to the search expression, each search means searches the database based on a partial search expression assigned to the search means by the division assigning means. Database search system to do. 3. A search target range setting means for setting a search target range common to each of the search means, wherein when the search target range is set, each of the search means is within the search target range. The database search system according to claim 1, which executes a search. 4. The database search system according to claim 1, wherein the plurality of search means include at least a search means for performing an index search and a search means for performing a full-text search.